skills/antfu/skills/pnpm/Gen Agent Trust Hub

pnpm

Pass

Audited by Gen Agent Trust Hub on Nov 5, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONPERSISTENCE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references official pnpm installation scripts hosted at get.pnpm.io and GitHub. These are well-known and trusted sources for provisioning the package manager in CI/CD workflows.- [REMOTE_CODE_EXECUTION]: As documented in the CI setup guide, the standard installation process involves downloading and executing a shell script from official pnpm repositories. This is a legitimate and standard procedure for the tool.- [INDIRECT_PROMPT_INJECTION]: The skill describes processing of project manifests (package.json) and workspace configurations (pnpm-workspace.yaml), which represent potential ingestion points for untrusted data. Ingestion points: package.json, pnpm-workspace.yaml, .pnpmfile.mjs. Boundary markers: None. Capability inventory: pnpm run, pnpm exec, pnpm install, pnpm rebuild (command execution). Sanitization: The skill thoroughly documents pnpm's allowBuilds security feature, which requires explicit user or configuration-based approval before executing third-party scripts.- [PERSISTENCE]: The skill mentions the pnpm setup command, which manages system-level persistence by modifying shell profiles (e.g., .bashrc, .zshrc) to add the tool to the environment PATH. This is standard, documented behavior for the package manager.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 5, 2026, 07:18 AM